Add options to print header keys for summary and fingerprinting

src/bin/fingerprint.rs

@@ -32,8 +32,12 @@ struct Args {
     #[arg(short, long, default_value_t = 10)]
     top_n: usize,
 
-    /// Include subpart params in fingerprint
+    /// Include subpart headers keys in fingerprint
     #[arg(short, long)]
+    include_header_keys: bool,
+
+    /// Include subpart params in fingerprint
+    #[arg(long)]
     include_params: bool,
 
     /// List of input directories to recursively search
@@ -72,7 +76,12 @@ fn main() -> anyhow::Result<()> {
             if should_skip(&arg) {
                 return None;
             }
-            match parse(&arg, youngest, args.include_params) {
+            match parse(
+                &arg,
+                youngest,
+                args.include_header_keys,
+                args.include_params,
+            ) {
                 Ok(Some(h)) => Some((h, arg)),
                 // Skip old emails
                 Ok(None) => return None,
@@ -123,6 +132,7 @@ fn main() -> anyhow::Result<()> {
 fn parse<P: AsRef<Path>>(
     path: P,
     youngest: i64,
+    include_header_keys: bool,
     include_params: bool,
 ) -> Result<Option<String>, EmailError> {
     let file = File::open(&path)?;
@@ -139,5 +149,7 @@ fn parse<P: AsRef<Path>>(
         return Ok(None);
     }
     //println!("{}: {:#?}", path.as_ref().display(), m.ctype);
-    Ok(Some(fingerprint(&m, include_params).join("\n")))
+    Ok(Some(
+        fingerprint(&m, include_header_keys, include_params).join("\n"),
+    ))
 }

src/bin/summarize.rs

@@ -9,8 +9,12 @@ use memmap::MmapOptions;
 #[derive(Parser, Debug)]
 #[command(author, version, about, long_about = None)]
 struct Args {
-    /// Include subpart params in fingerprint
+    /// Include subpart headers keys in fingerprint
     #[arg(short, long)]
+    include_header_keys: bool,
+
+    /// Include subpart params in fingerprint
+    #[arg(long)]
     include_params: bool,
 
     /// List of files to summarize
@@ -29,7 +33,7 @@ fn main() -> anyhow::Result<()> {
             .unwrap_or("(NO SUBJECT)".to_owned());
         println!(
             "{subject}: {path}\n{}",
-            fingerprint(&m, args.include_params).join("\n")
+            fingerprint(&m, args.include_header_keys, args.include_params).join("\n")
         );
     }
     Ok(())

src/lib.rs

@@ -65,17 +65,39 @@ pub fn should_skip(path: &str) -> bool {
     SKIP_FILES.contains(&filename)
 }
 
-pub fn fingerprint(pm: &ParsedMail<'_>, include_params: bool) -> Vec<String> {
+pub fn fingerprint(
-    fingerprint_rec(pm, include_params, 0)
+    pm: &ParsedMail<'_>,
+    include_header_keys: bool,
+    include_params: bool,
+) -> Vec<String> {
+    fingerprint_rec(pm, include_header_keys, include_params, 0)
 }
-fn fingerprint_rec(pm: &ParsedMail<'_>, include_params: bool, depth: usize) -> Vec<String> {
+fn fingerprint_rec(
+    pm: &ParsedMail<'_>,
+    include_header_keys: bool,
+    include_params: bool,
+    depth: usize,
+) -> Vec<String> {
     let indent = " ".repeat(depth * 2);
     let mut parts = vec![format!("{}{}", indent, pm.ctype.mimetype)];
-    for (k, v) in &pm.ctype.params {
+    if include_header_keys {
-        parts.push(format!("{indent} {k}: {v}"));
+        parts.push(format!("{indent}Headers:"));
+        for k in &pm.headers {
+            parts.push(format!("{indent} {}", k.get_key()));
+        }
+    }
+    if include_params {
+        for (k, v) in &pm.ctype.params {
+            parts.push(format!("{indent} {k}: {v}"));
+        }
     }
     for c in &pm.subparts {
-        parts.extend(fingerprint_rec(&c, include_params, depth + 1));
+        parts.extend(fingerprint_rec(
+            &c,
+            include_header_keys,
+            include_params,
+            depth + 1,
+        ));
     }
     parts
 }